The State Superintendent of Public Instruction appeals a decision holding that IQ tests used by the California school system to place children into special classes for the educable mentally retarded (E.M.R.) violated federal statutes and the equal protection clauses of the United States and California Constitutions. The district court enjoined the use of non-validated IQ tests, and ordered the state to develop plans to eliminate the disproportionate enrollment of black children in E.M.R. classes. We affirm on the statutory grounds and reverse on the federal and state constitutional issues.
I. PROCEDURE BELOW
The initial complaint for declaratory and injunctive relief was filed in 1971, with six black elementary schoolchildren in the San Francisco Unified School District as named plaintiffs. Appellees challenged as unconstitutional the use of standardized intelligence tests for placement of black children in E.M.R. classes in San Francisco. The defendants were the city and state superintendents, the members of the State Board of Public Instruction and the members of the City Board of Education. The district court certified the plaintiff class as consisting “of all black San Francisco schoolchildren who have been classified as mentally retarded on the bases of IQ test results” and granted appellees’ motion for a preliminary injunction. Larry P. v. Riles, 343 F.Supp. 1306 (N.D.Cal.1972). Defendants appealed from this judgment, which this court affirmed. Larry P. v. Riles, 502 F.2d 963 (9th Cir.1974).
In August 1973 the State defendants asked for a three-judge court, pursuant to 28 U.S.C. § 2281. The district court rejected this request. In December 1974 the district court expanded the class to include “all Black California school children who have been or may in the future be classified as mentally retarded on the basis of I.Q. tests.” The terms of the preliminary injunction were correspondingly expanded. In January 1975 defendants voluntarily placed a moratorium on all IQ testing for E.M.R. placement. In January and July of 1977 appellees filed amended complaints alleging violations of Title VI of the Civil Rights Act of 1964, 42 U.S.C. § 2000d et seq.; the Emergency School Aid Act of 1972 and 1974, 20 U.S.C. §§ 3191-3207 (repealed 1982); the Education For All Handicapped Children Act, 20 U.S.C. §§ 1401-1461; section 504 of the Rehabilitation Act of 1973, 29 U.S.C. § 794; the equal protection clauses of the United States and California Constitutions; and several sections of the California Education Code. Jurisdiction was predicated on 28 U.S.C. §§ 1331, 1337 and 1343 and 42 U.S.C. § 1983. In August 1977 the United States was granted permission to participate as amicus curiae.
Trial began on October 11,1977 and concluded on March 15, 1978. On October 16, 1979 the district court entered judgment for the appellees. The district court held that the plaintiffs were adequate representatives of the class of black children who had been or in the future would be wrongly placed and maintained in special classes for the educable mentally retarded. Larry P. v. Riles, 495 F.Supp. 926, 987 (N.D.Cal. 1979). The court held that the use of IQ tests for placement of black children in E.M.R. classes violated Title VI of the Civil Rights Act of 1964, 42 U.S.C. § 2000d et seq., the Rehabilitation Act of 1973, 29 U.S.C. § 794, and the Education For All Handicapped Children Act of 1975, 20 U.S.C. §§ 1401-1461, and the equal protection clauses of the federal and state constitutions. Larry P., 495 F.Supp. at 933. The district court found no violations of California statutory law. Id. at 986 n. 110.
The district court permanently enjoined the defendants from utilizing any standardized IQ test for the identification of black E.M.R. children or their placement into E.M.R. classes, without securing prior approval of the court. Id. at 989. The court ordered the defendants to direct each school district to re-evaluate every black child currently identified as an E.M.R. pupil *973without using standardized intelligence tests.
Further, the defendants were “ordered to monitor and eliminate disproportionate placement of black children in California’s E.M.R. classes.” Id. at 990. The district court specifically ordered the defendants to obtain an annual report from each school district on the racial proportions of E.M.R. classes, to prepare a statewide report, and to direct each school district with a black E.M.R. pupil enrollment one standard deviation above the district rate of white E.M.R. pupil enrollment to prepare a plan to correct the imbalance. The defendants were ordered to bring to the attention of the court any district imbalance if the disparity in excess of one standard deviation existed after three years.
Wilson Riles, California Superintendent of Public Instruction, filed a timely notice of appeal. The other defendants did not appeal.
II. FACTS
A detailed discussion of the development of IQ tests and of special education in California can be found in the thorough and well-reasoned district court opinion.1 Id. at 935-52. For the purposes of this opinion, we will summarize the district court’s findings of fact relevant to this appeal.
In the mid-60’s California created programs for several categories of students with educational problems. The “educable mentally retarded” (E.M.R.) program was for schoolchildren of retarded intellectual development who are considered incapable of being educated through the regular educational program, but who could benefit from special educational facilities to make them economically useful and socially adjusted. The “trainable mentally retarded” (T.M.R.) category was for children with more severe retardation than educable mentally retarded. CAL.EDUC.CODE § 56515 (West 1978) (repealed 1980). In addition, there were two categories for students who, with help, could be returned to a regular school program. These were the programs for “culturally disadvantaged minors,” children with cultural or economic disadvantages, but with potential for successfully completing a regular educational program, CAL.EDUC.CODE § 56600 (West 1978) (repealed 1980); and for “educationally handicapped minors” (E.H.), students with marked learning or behavioral disorders, capable of returning to a regular school program but who cannot presently benefit from the regular program. CAL. EDUC.CODE § 56600 (West 1978) (repealed 1980). Larry P., 495 F.Supp. at 937-38.
The E.M.R. classes are for children who are considered “incapable of learning in the regular classes,” and the E.M.R. curriculum “is not designed to help students learn the skills necessary to return to the regular instructional program.” Id. at 941 (emphasis in original). The E.M.R. classes are designed only to teach social adjustment and economic usefulness. Id.
“The [E.M.R.] classes are conceived of as ‘dead-end classes,’ ” and a misplacement in E.M.R. causes a stigma and irreparable injury to the student. Id. at 941-42.
From 1968 until trial in 1977, black children have been significantly overrepresented in E.M.R. classes. For example, in 1968-69, black children were about 9% of the state school population, yet accounted for 27% of the E.M.R. population. Id. at 938.
“These apparent overenrollments could not be the result of chance. For example, there is less than a one in a million chance that the overenrollment of black children and the underenrollment of non-black children in the E.M.R. classes in 1976-77 would have resulted under a color-blind system.” Id. at 944. To explain this over-enrollment, the defendants proffered a theory that there is a higher incidence of mental retardation among the black population. The district court found that this theory *974fails to account for the problem, because even “if it is assumed that black children have a 50 percent greater incidence of this type of mental retardation, there is still less than a one in 100,000 chance that the enrollment could be so skewed towards black children____ [Further,] the disproportionate E.M.R. enrollment of black children is not duplicated in the classes for the so-called ‘trainable mentally retarded’ children.” Id.
Since 1967 complaints and concern had been mounting about the use of IQ tests and the placement of minorities in E.M.R. classes. Id. at 939. In 1969 the state legislature enacted a resolution calling for a study by the State Board of Education of the overenrollment of minorities in E.M.R. classes. House Resolution 444. Litigation began in 1969 that raised many of these same issues on the behalf of children from Spanish-speaking backgrounds. Diana v. Board of Education, No. C-70-37 RFP (N.D.Cal.)2.
Prior to 1969, although the state required IQ tests and an individualized psychological examination prior to E.M.R. placement, there was no mandatory list of required IQ tests. In 1969 “the [State] Department [of Education (SDE) ] proposed and the State Board of Education adopted, an addition to the California Administrative Code requiring that approved IQ tests be used as part of the E.M.R. placement process.” Larry P., 495 F.Supp. at 946. The district court found that the SDE moved extremely quickly and unsystematically to select those IQ tests for the mandatory list. The district court further found that the person who oversaw this selection was not an expert in IQ testing, and that the SDE did not expressly consider or investigate the problems concerning the disproportionate enrollment of minorities or the cultural bias of IQ tests despite its awareness of these problems. In addition, the SDE contacted no independent testing experts regarding the compilation of the list, and ignored requests from field personnel to take more time to select tests. Id. at 946-47. The court concluded that “by relying on the most commonly used tests, [the Department] opted to perpetuate any discriminatory effects of those tests.” Id.
While the district court made the above findings to the effect that key officials in SDE were aware of the disproportionate enrollment of minority children in E.M.R. classes, and of widespread charges that cultural bias in the IQ tests caused or contributed to that disproportion, that did not necessarily apply to Superintendent Riles who is the only defendant who has appealed. That finding thus stands as to the nonappealing defendants. However, Dr. Riles did not become Superintendent of Public Instruction until 1971, whereas the E.M.R. programs were in effect much earlier. In mid-1969, Dr. Riles had served as head of the California Compensatory Education Programs and in that capacity was on record as telling the President’s Commission on Mental Retardation that, because of the disproportionate number of minorities in E.M.R. programs,
In California, educators are taking a second look at their classification criteria to see if language difficulties, deprivation of experiences, and deviation from the majority’s culture and value system may be entering into the determination of who is mentally retarded.
Id. at 945. The court found that state SDE’s actions, however, “revealed a studied effort to avoid the ‘second look’ suggested by Dr. Riles.” Id. at 946. But the court did not find that Dr. Riles joined in that avoidance.
In the findings Dr. Riles is clearly shown as disassociated from any suggestions of *975intentional discrimination in the selection, use, and validation of IQ tests as determinants for E.M.R. placement, id., or the acceptance of any theory of “genetic inferiority” as a justification for over representation of black children in E.M.R. classes. Id. at 955. The record shows this appellant’s recognition that IQ tests which had been adopted before he became Superintendent failed to eliminate cultural biases. While the findings tended to lump all state officials together, including the Superintendent, a discrete examination of the role actually taken by him, as reflected in the testimony and findings, does not show that affirmative intent to discriminate which is required under the fourteenth amendment.
In 1971 the California legislature made the following declaration:
The legislature hereby finds and declares that there should not be disproportionate enrollment of any socioeconomic, minority, or ethnic group pupils in classes for the mentally retarded and that the verbal portion of the intelligence tests which are utilized by some schools for such placement tends to underestimate the academic ability of such pupils.
CAL.EDUC.CODE § 56504 (West 1978) (repealed 1980). Further, the legislature required that
No pupil may be placed in a special education program for the mentally retarded unless a complete phychological [sic] examination by a credentialed school psychologist investigating such factors as developmental history, cultural background, and school achievement substantiates the retarded intellectual development indicated by the individual test scores. This examination shall include estimates of adaptive behavior [the ability to engage in social activities and perform everyday tasks].
CAL.EDUC.CODE § 56506 (West 1978) (repealed 1980).
The court contrasted the defendants’ testimony that IQ tests were not unduly relied upon for placement with other testimony presented at trial, and determined that the IQ tests were either determinative or pervasive in the placement process. Larry P., 495 F.Supp. at 949-50. The court noted SDE reports that showed that a large proportion of the pupil records revealed no “developmental history” or other information required to be utilized in the E.M.R. placement process, and that a number of districts admitted they made little or no use of adaptive behavior information for placement. Id. at 950.
The district court found that the requirement of parental consent for E.M.R. placement does not overcome any deficiencies caused by bias in the placement process, because “that consent is rarely withheld, particularly by minorities, since the mystique of teacher authority and IQ scores tends to overwhelm parents.” Id. at 950 n. 51.
Since the moratorium on IQ testing in 1975, the total percentage of black children in E.M.R. classes has not changed substantially. The district court, however, examined the data concerning new E.M.R. placements, which were made without IQ tests, and found that uncontradicted expert testimony showed that the four percent drop in the placement of black children into E.M.R. classes is not likely to have occurred by chance. Id. at 951-52.
On the average, black children score fifteen points, or one standard deviation, below white children on standardized intelligence tests. Thus, utilizing the premorato-rium criteria used by California for E.M.R. placement, “approximately two percent of the total population fall below the two standard deviation cut-off, while about 15 percent of black children fall below that level.” Id. at 954.
The court found that “the tests were never designed to eliminate cultural biases against black children; it was assumed in effect that black children were less ‘intelligent’ than whites____ The tests were standardized and developed on an all-white population, and naturally their scientific validity is questionable for culturally different groups.” Id. at 956-57. Since the 1920’s it has been generally known that black persons perform less well than white *976persons on the standardized intelligence tests. IQ tests had been standardized so that they yielded no bias because of sex. For example, when sample tests yielded different scores for boys and girls, the testing experts assumed such differences were unacceptable and modified the tests so that the curve in the standardization sample for boys and girls was identical. No such modifications on racial grounds has ever been tried by the testing companies. The district court noted that “the experts have from the beginning been willing to tolerate or even encourage tests that portray minorities, especially blacks, as intellectually inferior.” Id. at 955.
The district court analyzed and rejected the defendants’ arguments advanced at trial that would explain the test score differences, which theorized that the lower scores for blacks were the result of actual, relevant differences between black and white children. The first argument is the genetic argument, which states that natural selection has resulted in black persons having a “gene pool” with lower intelligence than whites. The district court found the assumptions underlying the genetic argument highly suspect, and in any event that the defendants “were unwilling to admit any reliance on [this theory] for policy-making purposes.” Id.
The second theory is the socioeconomic argument, which theorizes that because of blacks’ lower socioeconomic status, they are at a greater risk for all kinds of diseases due to malnutrition and poor medical attention. The district court found that the facts did not support this theory, since it did not explain why more severe mental retardation, e.g. that consistent with placement into classes for the trainable mentally retarded children, does not occur in greater proportions among blacks and poorer sections of the population. Id. at 956.
The district court found that the appellants failed to show that the IQ tests were validated for blacks with respect to the characteristics consistent with E.M.R. status and placement in E.M.R. classes, i.e., that the defendants failed to establish that the IQ tests were accurate predictors that black elementary schoolchildren who scored less than 70 were indeed mentally retarded. Id. at 971-73.
The district court found that alternatives to IQ testing for E.M.R. placement have been in effect since the state moratorium on IQ testing in 1975. These procedures, in which schools take more time and care with their assessments for E.M.R. classification and rely more on observational data, are less discriminatory than under the IQ-centered standard. Id. at 973-74.
The district court found that defendants were guilty of intentional discrimination in the use of the IQ tests for E.M.R. placement. The court based this determination on the facts that the historical background of the IQ tests shows cultural bias; the adoption of the mandatory IQ testing requirement in 1969 was riddled with procedural and substantive irregularities, in which no outside sources were consulted by the State Board and the question of bias was never considered, even though the officials were well aware of the bias and disproportionate placement problems caused by the IQ tests (this problem having been addressed in a legislative resolution); the defendants’ “complete failure to ascertain or attempt to ascertain the validity of the tests for minority children;” and the failure of the state to investigate and act on legal requirements to report significant variances in racial and ethnic composition in E.M.R. classes. Id. at 980-82. The court noted that “the SDE’s actions revealed a complacent acceptance of those disproportions, and that complacency was evidently built on easy but unsubstantiated assumptions about the incidence of retardation or at least low intelligence among black children.” Id. at 983.
The court found that the named plaintiffs’ assignments to E.M.R. classes were based on their scores on IQ tests, and that these plaintiffs were either in E.M.R. classes or subject to reassignment at the time the class was certified. Id. at 987. Further, the court found that “the only *977relevant evidence on their cases indicated that they were not retarded.” Id. at 988.
III. ISSUES
(1) Whether this case is moot because the named plaintiffs were not adequate class representatives at the time the class was certified because (a) IQ tests were not involved in the named plaintiffs’ placement in the E.M.R. classes; or because (b) none of the named plaintiffs were placed in an E.M.R. class after statutory changes in the placement procedure became effective in October 1971.
(2) Whether the relief granted required a three-judge panel pursuant to 28 U.S.C. § 2281.
(3) Whether the district court erred in holding that appellant’s use of the IQ tests violated the Rehabilitation Act of 1973 and the Education For All Handicapped Children Act of 1975, because the court clearly erred in finding that appellant had failed to show the tests were validated for black children and that appellant had failed to use a variety of statutorily mandated evaluation tools.
(4) Whether the district court erred in holding that appellant’s action violated Title VI, because the court clearly erred in finding that the tests were not validated, that placement in E.M.R. classes for non-E.M.R. students was not a benefit, and that appellant failed to show any countervailing reason why a disproportionate number of black children were placed in the E.M.R. classes.3
(5) Whether the district court erred in holding appellant’s action violated the equal protection clause of the CAL. Const, art. I, § 7.
(6) Whether the remedy directing alleviation of disproportionate placement of black children in E.M.R. classes is overbroad.
IV. MOOTNESS
The class, originally certified in June 1972 and modified in November 1974, is defined as “the class of all black California school children who have been or may in the future be classified as mentally retarded on the basis of IQ tests.”
Appellant argues that since none of the named plaintiffs were placed in E.M.R. classes immediately after taking an IQ test, their placement was not made in reliance on the IQ tests. The district court, however, made a finding that the individual plaintiffs were assigned to E.M.R. classes based notably on the use of IQ tests. Larry P., 495 F.Supp. at 987. There was substantial testimony by witnesses for both sides that the IQ score was a main factor in placement in E.M.R. classes. Further, a report by the California Association of School Psychologists and Psychometrists (CASPP), prepared at the request of the appellant in preparation for trial, stated that the official school records contained insufficient evidence of utilization of educational history, health and developmental history or of social and cultural data, and that the “thoroughness of examination was not adequate for placement.”
In addition, the district court found that the only relevant evidence on the named plaintiffs indicated they were not retarded. Id. at 988. This finding itself is not clearly erroneous: the CASPP report concluded that none of the plaintiffs belonged in E.M.R. classes, and none of the personnel who tested, evaluated or taught the plaintiffs testified that the children were mentally retarded. Only one of appellant’s witnesses testified that the named plaintiffs *978were mentally retarded, and this witness never personally interviewed the named plaintiffs, and was not aware that other school psychologists who had evaluated the plaintiffs had concluded that they were not mentally retarded.
Based on the evidence that the named plaintiffs were not mentally retarded, that their school records lacked other data that should have been used for placing them in E.M.R. classes, and that the IQ score was a main factor in placement, the district court’s finding that IQ tests were notably used in their placement in E.M.R. classes is not clearly erroneous, notwithstanding the length of time from testing to placement.
Appellant next argues that none of the named plaintiffs are adequate class representatives since none of them were placed in E.M.R. classes after legislative changes in the placement procedure became effective October 1, 1971. The procedure changed in two important respects: (1) written consent of the parent must now be given for placement; and (2) a complete individual psychological examination by a credentialed school psychologist is required, investigating developmental history, cultural background, school achievement and estimates of adaptive behavior.
The district court made two findings that addressed this argument which appellant has not shown to be clearly erroneous. First, the district court found that parental “consent is rarely withheld, particularly by minorities, since the mystique of teacher authority and IQ scores tends to overwhelm parents.” Id. at 950 n. 51. This finding is supported by testimony from experts in special education, as well as by testimony from employees that illustrated that parental consent had been requested by the school authorities prior to, rather than after, an admissions committee hearing which evaluated the students, and that often the consent was obtained by an employee who did not have specific information about the students. Appellant points to no testimony indicating that this finding is clearly erroneous, or that the new requirement of parental consent overcame any deficiencies caused by bias in the placement process.
The district court made a finding that “the entire placement process revolves around the IQ determination,” that the IQ scores were very pervasive in the placement process, and that this continued until 1975 when the court issued an injunction against the use of IQ tests. Id. at 950.
Based on the evidence in the record, we cannot say the district court was clearly erroneous in inferring that the statutory changes in procedures effective in 1971 did not change the prominence of the use of IQ tests for E.M.R. placement. Therefore, the fact that the named plaintiffs were placed in E.M.R. classes prior to the 1971 changes does not moot the case.
V. THREE-JUDGE COURT
28 U.S.C. § 2281 (1965) (repealed 1976) required a three-judge court to hear actions seeking an injunction against the enforcement or operation of state statutes on grounds of unconstitutionality.4
Appellant claims that the effect of the district court’s orders in this case is to enjoin the operation of CAL.EDUC.CODE § 56505 (West 1978) (repealed 1980) on constitutional grounds, and thus a three-judge court was required.
CAL.EDUC.CODE § 56505 (West 1978) (repealed 1980) states that “[bjefore any *979pupil is admitted to a special education program for mentally retarded pupils ... the pupil shall be given verbal or nonverbal individual intelligence tests____ Such tests shall be selected from a list approved by the State Board of Education.” Moreover, CAL.EDUC.CODE § 56504 (West 1978) (repealed 1980) “declares that there should not be disproportionate enrollment of any socioeconomic, minority, or ethnic group pupils in the classes for the mentally retarded and that the verbal portion of the intelligence tests which are utilized by some schools for such placement tends to underestimate the academic ability of such pupils.”
It is clear that the intent of the California legislature from these two statutes construed together, while permitting the use of intelligence tests prior to placement in E.M.R. classes, is to require that the tests selected by the State Board of Education should not be culturally biased against racial minorities. California statutes do not require administration of biased or not validated IQ tests.
The gravamen of appellees’ complaint is not against the use of IQ tests in general, but that the particular tests selected by the State Board of Education were culturally or racially biased. Appellees therefore are not attacking the constitutionality of a state statute; they merely challenge the actions of the State Board of Education in attempting to comply with the statute. A three-judge court is not required when “seeking an injunction on the ground of the unconstitutionality of the result obtained by the use of a statute not attacked as unconstitutional.” Turner v. Fouche, 396 U.S. 346, 353 n. 10, 90 S.Ct. 532, 536 n. 10, 24 L.Ed.2d 532 (1970). See also Safeguard Mutual Insurance Co. v. Pennsylvania, 329 F.Supp. 315, 318 (E.D. Pa.1971).
VI. REHABILITATION ACT
The Rehabilitation Act of 1973 provides inter alia:
No otherwise qualified handicapped individual in the United States, as defined in section 706(7) of this Title, shall, solely by reason of his handicap, be excluded from the participation in, be denied the benefits of, or be subjected to discrimination under any program or activity receiving Federal financial assistance.
29 U.S.C. § 794.5
The Education For All Handicapped Children Act (“EAHCA”), 20 U.S.C. §§ 1401-1461, provides that a state qualifying for federal assistance under the Act must establish:
procedures to assure that testing and evaluation materials and procedures utilized for the purposes of evaluation and placement of handicapped children will be selected and administered so as not to be racially or culturally discriminatory ... [N]o single procedure shall be the sole criterion for determining an appropriate educational program for a child.
20 U.S.C. § 1412(5)(C).
Congress was clearly concerned with the misclassification of students as retarded. The Senate Report for the Rehabilitation Act states that “racial and ethnic factors may contribute to misclassification as mentally retarded.” S.Rep. No. 93-1297, 93d Cong., 2d Sess. reprinted in 1974 U.S. CODE CONG. & AD.NEWS 6373, 6389. The Senate Report for the Education For All Handicapped Children Act states that “[t]he [Labor and Public Welfare] Committee is deeply concerned about practices and procedures which result in classifying children as having handicapping conditions when, in fact, they do not have such conditions.” S.Rep. No. 94-168, 94th Cong., 1st Sess. 26, reprinted in 1975 U.S.CODE CONG. & AD.NEWS 1425, 1450.
The Department of Education adopted almost identical regulations under both of these acts. These regulations require that recipients of federal funds under the acts ensure that:
*980[t]ests and other evaluation materials have been validated for the specific purpose for which they are used ... [and] assess specific areas of educational need and not merely those which are designed to provide a single general intelligence quotient.
34 C.F.R. § 104.35(b)(1), (2) and 34 C.F.R. § 300.532(a)(2), (b), originally adopted as 45 C.F.R. § 121a.532(a)(2), (b).
The regulations further provide that evaluation and placement be based on “a variety of sources, including aptitude and achievement tests, teacher recommendations, physical conditions, social or cultural background, and adaptive behavior.” 34 C.F.R. § 104.35(c)(1), (2); 34 C.F.R. § 300.-533(a)(1), originally adopted as 45 C.F.R. § 121a.533(a)(l), (2). Placement is required to not be based upon a single criterion. 20 U.S.C. § 1412(5)(C); 34 C.F.R. § 300.532(d), originally adopted as 45 C.F.R. § 121a.532(d).
In summary, the Education For All Handicapped Children Act specifically requires that tests and evaluation procedures be free of racial and cultural bias. Both the EAHCA and the Rehabilitation Act require that the tests used for evaluation be validated for the specific purpose for which they are used, and that placement not be based upon a single criterion but on a variety of sources. It is undisputed that California receives substantial federal assistance under these acts, and thus is required to conform to these procedures and regulations. The regulations place the burden on the recipient to show it has complied with the requirements.
Appellant argues that the IQ tests were validated for the specific purposes for which they are used. Appellant analogizes to Title VII cases, notably Washington v. Davis, 426 U.S. 229, 96 S.Ct. 2040, 48 L.Ed.2d 597 (1976), for the proposition that tests that are valid predictors of future performance can be utilized even if they have a discriminatory impact. There are two problems with appellant’s proposition. First, the employment context is quite different from the educational situation. As the district court stated, “[i]f tests can predict that a person is going to be a poor employee, the employer can legitimately deny that person a job, but if tests suggest that a young child is probably going to be a poor student, the school cannot on that basis alone deny that child the opportunity to improve and develop the academic skills necessary to success in our society.” Larry P., 495 F.Supp. at 969. Assigning a student to an E.M.R. class denies that child the opportunity to develop the necessary academic skills, since E.M.R. classes do not teach academic subjects and are essentially a dead-end academic track. Second, and more important, the question for predictive validity in schools is not whether the standardized intelligence tests predict future school performance generally, as appellant argues, but whether the tests predict specifically that black elementary schoolchildren (as opposed to white elementary schoolchildren) who score at or below 70 on the IQ tests are mentally retarded and incapable of learning the regular school curriculum. In this case, the appellant would have to have shown that the tests are a proven tool to determine which students have characteristics consistent with E.M.R. status and placement in E.M.R. classes, i.e., “whose mental capabilities make it impossible for them to profit from the regular educational programs” even with remedial instruction. The regulations place the burden of showing such validation on the defendants.
The district court found that defendants failed to show that the tests were validated for placing black students with scores of 70 or less in E.M.R. classes. The district court noted that very few studies had examined the difference of IQ predictability for black as compared to white populations, and that those studies which had examined this problem found the tests much less valid for blacks than for whites. Further, the district court found that, even assuming the tests were validated for placement of white schoolchildren in E.M.R. classes, such validation for blacks had been generally assumed but not estab*981lished. For example, the tests had been adjusted to eliminate differences in the average scores between the sexes, but such adjustment was never made to adjust the scores to be equal for black and white children. Id. at 971. The court found that the reason for this was a basic assumption of a lower level of intelligence in blacks than in whites. The fact that early test developers indeed made this assumption is borne out by the literature and testimony at trial. See id. at 954-60. In addition, no studies have been made, either by the defendants or the testing companies, to investigate the reasons for the one standard deviation difference in test scores between the races or to determine whether test redesign could eliminate any bias. Id. at 971. There was expert testimony that a much larger percentage of black than white children had been misplaced in E.M.R. classes. Based on the evidence in the record, the district court finding that the appellant had not established validation of the test is not clearly erroneous.6
The district court also found that the appellant did not utilize the variety of information required by statute and regulation to make E.M.R. placements, but relied primarily on the IQ test. This finding also is not clearly erroneous. Testimony showed that school records lacked sufficient evidence of educational history, adaptive behavior, social and cultural background or health history for these factors to have been utilized in placement. See IV, supra.
Since the appellant has not shown that these findings are clearly erroneous, we affirm the district court’s holding that the defendants violated the provisions of the Rehabilitation Act and the Education For All Handicapped Children Act (1) by not insuring that the tests were validated for the specific purpose for which they are used, and (2) by not using the variety of statutorily mandated evaluation tools.
VII. TITLE VI
Title VI of the Civil Rights Act of 1964, 42 U.S.C. § 2000d, provides that:
No person in the United States shall, on the ground of race, color, or national origin, be excluded from participation in, be denied the benefits of, or be subjected to discrimination under any program or activity receiving federal financial assistance.
Regulations issued under this statutory mandate require that recipients of federal funding may not
utilize criteria or methods of administration which have the effect of subjecting individuals to discrimination because of their race, color, or national origin, or have the effect of defeating or substantially impairing accomplishment of the objectives of the program as respect individuals of a particular race, color, or national origin.
34 C.F.R. 100.3(b)(2), originally adopted as 45 C.F.R. § 80.3(b)(2).
In Guardians Association v. Civil Service Commission of City of New York, 463 U.S. 582, 103 S.Ct. 3221, 77 L.Ed.2d 866 (1983), a majority of the Court held that a violation of Title VI required proof of discriminatory intent.7 A different majority held, however, that proof of discriminatory effect suffices to establish liability when *982the suit is brought to enforce regulations issued pursuant to the statute rather than the statute itself.8
The appellees relied on the regulations issued pursuant to Title VI. Larry P., 495 F.Supp. at 965. The lower court held that the placement mechanisms for E.M.R. classes operated with a discriminatory effect in violation of the regulations and HEW’s “interpretative guidelines”. Id. In light of appellees’ reliance on the regulations, we find it appropriate to apply a discriminatory effect analysis.
A prima facie case is demonstrated by showing that the tests have a discriminatory impact on black schoolchildren. Board of Education of New York v. Harris, 444 U.S. 130, 151, 100 S.Ct. 363, 375, 62 L.Ed.2d 275 (1979). Once a plaintiff has established a prima facie case, the burden then shifts to the defendant to demonstrate that the requirement which caused the disproportionate impact was required by educational necessity.9 Debra P. v. Turlington, 644 F.2d 397, 407 (5th Cir. 1981). See also Connecticut v. Teal, 457 U.S. 440, 446-47, 102 S.Ct. 2525, 2530-31, 73 L.Ed.2d 130 (1982) (Title VII); Coe v. Yellow Freight System, Inc., 646 F.2d 444, 451 (10th Cir.1981) (Title VII).10
Appellees clearly demonstrated the discriminatory impact of the challenged *983tests. It is undisputed that black children as a whole scored ten points lower than white children on the tests, and that the percentage of black children in E.M.R. classes was much higher than for whites. As discussed previously, these test scores were used to place black schoolchildren in E.M.R. classes and to remove them from the regular educational program. The burden therefore shifted to the defendants to demonstrate that the IQ tests which resulted in the disproportionate placement of black children were required by educational necessity.
Appellant argues first that E.M.R. classes are a benefit for, rather than adverse discrimination against, black children, implying that appellees did not even establish a prima facie case. However, the district court found that improper placement in E.M.R. classes has a definite adverse effect, in that E.M.R. classes are dead-end classes which de-emphasize academic skills and stigmatize children improperly placed in them. Larry P., 495 F.Supp. at 941-42. Even appellant’s witnesses testified that it would be extremely improper for a non-mentally retarded child to be placed in an E.M.R. classroom. Though the E.M.R. class might be a benefit for those students who are educable mentally retarded, it is clearly damaging to a non-retarded student to be placed in those classes. The district court’s finding is not clearly erroneous, and thus appellees established a prima facie case of a Title VI violation based upon discriminatory effect.
Appellant next argues that even if the impact is adverse, it is not caused by discriminatory criteria (the IQ tests), but by other nondiscriminatory factors: (1) placement is based on a variety of information and evaluation tools that are non-discriminatory, and not solely on the IQ tests; (2) the tests are validated for black schoolchildren, and therefore accurately reflect mental retardation in black children; and (3) blacks have a higher percentage of mental retardation than whites.
Appellant’s first two arguments have been discussed in VI, supra, and are unavailing. Appellant’s third argument is that the disproportionate number of black children in E.M.R. classes is based on a higher incidence of mental retardation in blacks than in whites that is due to poor nutrition and poor medical care brought on by the lower socioeconomic status of blacks. This argument also fails. Appellees showed, and the district court made a finding, that “the overrepresentation of black children in E.M.R. classes cannot be explained away solely on the grounds of the generally lower socio-economic status of black children and their parents.” Id. at 956. The district court specifically found the testimony of appellant’s experts in support of this argument failed to explain why more severe mental retardation does not occur in greater proportions among the poorer sections of the population. In addition, there was testimony from other experts that poor nutrition or medical care during early life does not affect later performance on IQ scores, unless it is a severe malnutrition of a type that is rare in this country. This finding of the district court has not been shown to be clearly erroneous.
Because appellant has failed to show that any of these district court findings were clearly erroneous, the district court did not err in holding that the defendants violated Title VI by utilizing these IQ tests for placement into E.M.R. classes.
VIII. EQUAL PROTECTION VIOLATION
We have previously held that a violation of Title VI is sustained under the “discriminatory effect” analysis when deal*984ing with the placement mechanisms for E.M.R. classes, partially in the light of the Supreme Court’s decision in Guardians Association. We cannot, however, sustain the finding of a violation by Superintendent Riles of the equal protection clause of the fourteenth amendment on the theory that the pervasiveness of discriminatory effect can, without more, be equated with the discriminatory intent required by Washington v. Davis. Accordingly, we reject these facts of the trial* court and reverse the conclusions that the Superintendent was guilty of intentional discrimination under the fourteenth amendment.
IX. CALIFORNIA CONSTITUTION
The district court found that plaintiffs proved a violation of Article I, § 7, California’s constitutional equal protection provision. In Pennhurst State School and Hospital v. Halderman, 465 U.S. 89, 104 S.Ct. 900, 919, 79 L.Ed.2d 67 (1984), the Court has held that federal courts lack jurisdiction to adjudicate cases against states based upon claims of violation of state constitutional or statutory law even when such cases are brought under pendant jurisdiction.
X. REMEDY
The district court, in addition to enjoining the use of nonvalidated intelligence tests and requiring re-evaluation of every current black E.M.R. pupil, ordered the defendants to require every school district that had a racial disproportion in E.M.R. classes to devise a three-year remedial plan, and to bring to the court’s attention any disparities that persist at the end of this period. Larry P., 495 F.Supp. at 989-90.
Appellant argues that, since appellees’ challenge was solely to the use of IQ tests, the only remedy within the power of the court was the elimination of the tests. Appellees’ prayer for relief, however, included a request for an injunction against defendants from retaining any black children presently enrolled in E.M.R. classes who are not mentally retarded, and to place these students in regular classrooms. Since the court found that black students had been improperly placed in E.M.R. classes, it is clearly within its power to correct this error.
Appellant next contends that the court’s order to eliminate the disproportionate enrollment requires an impermissible quota. The district court’s order, though, does not impose any quota on future E.M.R. placements. Even with respect to eliminating present disproportionate enrollment, the district court allowed an error leeway of one standard deviation. Further, the district court does not require that students properly placed in an E.M.R. class be removed. The court only required that any disproportion in excess of one standard deviation after three years be brought to the court’s attention. If the school district can show, utilizing properly validated procedures complying with applicable statutes, that the black students in the E.M.R. classes properly belong there, there would be no need to eliminate the remaining disproportion. Since there are no fixed numerical requirements, the district court’s order is not an impermissible quota. See Regents of the University of California v. Bakke, 438 U.S. 265, 98 S.Ct. 2733, 57 L.Ed.2d 750 (1978).
CONCLUSION
The named plaintiffs were proper class representatives. A three-judge district court pursuant to 28 U.S.C. § 2281 was not required. The district court properly concluded that appellant violated federal statutes. We reverse, however, the court’s finding of a federal constitutional violation. The lower court was without jurisdiction to review the state constitutional claims, and accordingly we vacate that portion of the decision below.
The judgment of the district court is AFFIRMED IN PART AND REVERSED IN PART. Costs are awarded to plaintiffs-appellees.
. We commend Judge Peckham for a detailed and comprehensive opinion and for his detailed summarizing and analyzing of the issues of a very complex and protracted trial.
. Diana made essentially the same claims as Larry P. with respect to Hispanic children, and also involved the issue of linguistic discrimination. The issues of that case were resolved through a stipulated settlement approved by the court on June 18, 1973. Under the terms of the settlement, the State Board of Education agreed to inquire into any school district in which there was a "significant variance" between the expected and actual percentage of Mexican-Americans in E.M.R. classes, to collect relevant data and to develop a plan to eliminate any such variances or disproportionate enrollments.
. Appellant does not argue that there is no private right of action under these federal statutes. Congress expressly recognized a private right of action to enforce the requirements of the EAH-CA. See 20 U.S.C. § 1415(e)(2), 34 C.F.R. § 121a.511, formerly 45 C.F.R. § 121a.511; Mt. View-Los Altos Union High School District v. Sharron B.H., 709 F.2d 28, 29 (9th Cir.1983) (EAHCA grants a private right of action). Private rights of action are recognized under Title VI, see, e.g., Lau v. Nichols, 414 U.S. 563, 94 S.Ct. 786, 39 L.Ed.2d 1 (1974); see also Cannon v. University of Chicago, 441 U.S. 677, 99 S.Ct. 1946, 60 L.Ed.2d 560 (1979) (Title IX), and under the Rehabilitation Act. see Lloyd v. Regional Transportation Authority, 548 F.2d 1277, 1285 (7th Cir.1977); Leary v. Crapsey, 566 F.2d 863, 865 (2d Cir.1977).
. 28 U.S.C. § 2281 provided:
“An interlocutory or permanent injunction restraining the enforcement, operation or execution of any State statute by restraining the action of any officer of such State in the enforcement or execution of such statute or of an order made by an administrative board or commission acting under State statutes, shall not be granted by any district court or judge thereof upon the ground of the unconstitutionality of such statute unless the application therefor is heard and determined by a district court of three judges under section 2284 of this title."
This statute was repealed in 1976. Pub.L. 94-381. The repeal does not apply to actions like the present case commenced on or before August 12, 1976. Id., § 7 (1976).
. Even though the named plaintiffs claim not to be handicapped (retarded), they are still protected by this Act. The Rehabilitation Act provides that it covers persons who are "regarded” as handicapped. 29 U.S.C. § 706(7)(B); 29 C.F.R. § 32.3.
. Appellant also argues that the failure of the district court to find that the tests were biased precludes any relief for appellees. This argument, however, misstates the burden of proof under the Rehabilitation Act and the EAHCA. The regulations under these acts place the burden on the defendant to show validation. In a Title VI case, once a prima facie case has been established, the burden is on the defendant to demonstrate that the criteria used are required by educational necessity. See note 8, infra. The defendants have failed to carry their burden.
. The Court previously held that no discriminatory intent must be proved to show a violation of the statute. Lau v. Nichols, 414 U.S. 563, 568, 94 S.Ct. 786, 789, 39 L.Ed.2d 1 (1974). This circuit has expressly followed Lau. Guadalupe Organization v. Tempe Elementary School, 587 F.2d 1022, 1029 (9th Cir. 1978); De La Cruz v. Tormey, 582 F.2d 45, 61, n. 16 (9th Cir.1978), cert. denied, 441 U.S. 965, 99 S.Ct. 2416, 60 L.Ed.2d 1072 (1979). The district court applied Lau while recognizing that subsequent dicta in Supreme Court opinions questioned that conclusion. E.g., Board of Education of New York v. Harris, 444 U.S. 130, 159-60, 100 S.Ct. 363, 379-*98240, 62 L.Ed.2d 275 (1979) (dissenting opinion of Stewart, Powell, and Rehnquist); Regents of the University of California v. Bakke, 438 U.S. 265, 284-87, 98 S.Ct. 2733, 2745-46, 57 L.Ed.2d 750 (1978). Nevertheless, even after Bakke, the Supreme Court reaffirmed an "effects" standard. Fullilove v. Klutznick, 448 U.S. 448, 479-80, 100 S.Ct. 2758, 2775-76, 65 L.Ed.2d 902 (1980). Despite these conflicting signals, a majority of Justices now agree that discriminatory intent is required. Guardians Association, 103 S.Ct. at 3236-37 (Powell, J. and Burger, C.J.); 103 S.Ct. at 3237 (Rehnquist, J., concurring in the judgment); 103 S.Ct. at 3237-38 (O'Connor, J., concurring); 103 S.Ct. at 3253 (Stevens, J., dissenting and joined by Brennan and Blackmun, JJ.).
. 103 S.Ct. at 3235, n. 1 (Powell, J., concurring) (listing and analyzing votes).
. This analysis derives from that applied in Title VII disparate impact claims. In the Title VII situation, "[t]o establish a prima facie case of discrimination, a plaintiff must show that the facially neutral employment practice had a significantly discriminatory impact. If that showing is made, the employer must then demonstrate that ‘any given requirement [has] a manifest relationship to the employment in question,’ in order to avoid a finding of discrimination.” Connecticut v. Teal, 457 U.S. 440, 446-47, 102 S.Ct. 2525, 2530-31, 73 L.Ed.2d 130 (1982) (quoting Griggs v. Duke Power Co., 401 U.S. 424, 432, 91 S.Ct. 849, 854, 28 L.Ed.2d 158 (1971)); Do-thard v. Rawlinson, 433 U.S. 321, 329, 97 S.Ct. 2720, 2726, 53 L.Ed.2d 786 (1977); Moore v. Hughes Helicopter, 708 F.2d 475, 481 (9th Cir. 1983). In the Title VI disproportionate impact claim in the educational situation, the defendant must therefore show that any given requirement has a manifest relationship to the education in question, i.e., that the IQ tests are required by "educational necessity.” Board of Education of New York v. Harris, 444 U.S. at 151, 100 S.Ct. at 375.
. Although in Title VII disparate treatment cases the burden of proof always remains with the plaintiffs, Texas Dept. of Community Affairs v. Burdine, 450 U.S. 248, 252-56, 101 S.Ct. 1089, 1093-95, 67 L.Ed.2d 207 (1981), this is not true in Title VI cases. In a Title VII disparate treatment case, the plaintiff has the initial burden of proving a prima facie case of discrimination. The burden of production then shifts to the defendant to articulate some legitimate non-discriminatory reason for the employee’s rejection. If the defendant carries this burden, the plaintiff then has an opportunity to prove by a preponderance of the evidence that the legitimate reasons offered by the defendant were a pretext for discrimination. The plaintiff always has the burden of persuasion. In addition, to prove a disparate treatment Title VII violation, the plaintiff must prove that the defendant intentionally discriminated against the plaintiff. Id.
"While Burdine thus affects the burdens of the parties in Title VII disparate treatment cases, it does not address Title VII disparate impact cases," Johnson v. Uncle Ben’s, Inc., 657 F.2d 750, 752 (5th Cir.1981), cert. denied, 459 U.S. 967, 103 S.Ct. 293, 74 L.Ed.2d 277 (1982) nor Title VI cases. In disproportionate impact cases under Titles VI and VII, no question of intentional discrimination arises. Teamsters v. United States, 431 U.S. 324, 335 n. 15, 97 S.Ct. 1843, 1854 n. 15, 52 L.Ed.2d 396 (1977); Lau, 414 U.S. at 568, 94 S.Ct. at 789. The Supreme Court in Burdine noted that disproportionate impact cases utilized different types of evidence than disparate treatment cases. Burdine, 450 U.S. at 252, 101 S.Ct. at 1093 n. 5. The “allocative scheme in Burdine was not formulated to develop the factual inquiry in a disparate impact case,” Johnson v. Uncle Ben’s, Inc., 657 F.2d at 752 n. 2, and thus Burdine does not affect prior precedent in Title VI and Title VII cases, which places the burden on the defendant to prove that the challenged action was required by em*983ployment or educational necessity. Contreras v. City of Los Angeles, 656 F.2d 1267, 1275 n. 5 (9th Cir.1981), cert. denied, 455 U.S. 1021, 102 S.Ct. 1719, 72 L.Ed.2d 140 (1982); Johnson v. Uncle Ben's, Inc., 657 F.2d at 753 n. 3; Coe v. Yellow Freight Systems, Inc., 646 F.2d at 451; Teal v. Connecticut, 645 F.2d 133, 136 n. 5 (2d Cir. 1981), aff'd, 457 U.S. 440, 102 S.Ct. 2525, 73 L.Ed.2d 130 (1982); Vuyanich v. Republic National Bank, 521 F.Supp. 656 (N.D.Tex.1981); Heffernan v. Western Electric Co., 510 F.Supp. 712, 715-16 (N.D.Ga.1981) (all Title VII cases).